Climate change impacts and adaptation options for the greek agriculture in 2021 2050: a monetary assessment

Climate change impacts and adaptation options for the Greek agriculture in 2021 2050 A monetary assessment Accepted Manuscript Climate change impacts and adaptation options for the Greek agriculture i[.]

Accepted Manuscript

Climate change impacts and adaptation options for the Greek agriculture in

2021-2050: A monetary assessment

E Georgopoulou, S Mirasgedis, Y Sarafidis, M Vitaliotou, D.P Lalas, I

Theloudis, K.-D Giannoulaki, D Dimopoulos, V Zavras

Please cite this article as: E Georgopoulou, S Mirasgedis, Y Sarafidis, M Vitaliotou, D.P Lalas, I Theloudis,

K.-D Giannoulaki, K.-D Dimopoulos, V Zavras, Climate change impacts and adaptation options for the Greek agriculture

in 2021-2050: A monetary assessment, Climate Risk Management (2017), doi: http://dx.doi.org/10.1016/j.crm.2017.02.002

This is a PDF file of an unedited manuscript that has been accepted for publication As a service to our customers

we are providing this early version of the manuscript The manuscript will undergo copyediting, typesetting, andreview of the resulting proof before it is published in its final form Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain

CLIMATE CHANGE IMPACTS AND ADAPTATION OPTIONS FOR THE GREEK

AGRICULTURE IN 2021-2050: A MONETARY ASSESSMENT

E Georgopoulou a, *, S Mirasgedis a , Y Sarafidis a , M Vitaliotou b , D P Lalas b

I Theloudis c , K.-D Giannoulaki c , D Dimopoulos c , V Zavras c

a National Observatory of Athens/IERSD, I Metaxa & Vas Pavlou, GR-15236 Palea Penteli, Greece

b FACE3TS S.A., 1 Agiou Isidorou str., GR-11471 Athens, Greece

c Piraeus Bank S.A., Environment Unit, 4 Amerikis str., GR-10564 Athens, Greece

* Corresponding author Tel.: +30 210 8109215, elenag@noa.gr

E-mail addresses: elenag@noa.gr, seba@noa.gr, sara@noa.gr, mabita@otenet.gr, lalas@facets.gr,

TheloudisI@piraeusbank.gr, GiannoulakiK@piraeusbank.gr, DimopoulosD@piraeusbank.gr, v.zavras@piraeusbank.gr

Abstract: The paper presents a quantitative assessment of mid-term (2021-2050) climate change impacts

on and potential adaptation options for selected crops in Greece that are of importance in terms of their share in national agricultural production and gross value added Central points in the assessment are the monetary evaluation of impacts and the cost-benefit analysis of adaptation options To address local variability in current and future climate conditions, analysis is spatially disaggregated into geographical regions using as an input downscaled results from climatic models For some crops (cereals, vegetables, pulses, grapevines), changes in future agricultural yields are assessed by means of agronomic simulation models, while for the rest crops changes are assessed through regression models The expected effects on crop yields of a number of potential adaptation options are also investigated through the same models, and the costs and benefits of these options are also quantitatively assessed The findings indicate that climate change may create winners and losers depending on their agricultural activity and location, while adaptation can mitigate adverse effects of climate change under cost-effective terms

Keywords: climate change; agriculture; impacts; adaptation; economic assessment

1 Introduction

Agriculture is one of the major economic sectors where climate change can have large impacts, affecting crop growth and consequently productivity As agricultural activities ensure food supply and represent an important source of income for local economies, especially in southern Europe, the investigation of these impacts is particularly important as it can provide the necessary scientific input for proper planning of adaptation strategies Especially in the Mediterranean region where Greece is located (as well as in the rest southern Europe), the recent findings of the Intergovernmental Panel of Climate Change (IPCC) reveal that the duration and intensity of droughts -as projected by regional and global climate models- will increase and will be accompanied by significant reductions in summer soil moisture (Kovats et al., 2014) These, together with temperature increase, entail dangers for crop cultivations

Agriculture in Greece is important both at national and regional level (as is also the case in other southern European countries like Italy, Spain and Portugal) In 2015, agriculture generated 4% of the Greek gross value added, while its share in some regions is even higher (of the order of 7-10%) In the case of Greece, agriculture is viewed, together with tourism, is viewed as a sector whose development will contribute substantially in providing the development that the local economy needs to mitigate the financial difficulties that have plagued it in the last years

To date very few studies have attempted to provide quantitative estimates of the climate change impacts

on crop cultivations in Greece together with the expected economic effects of adaptation Giannakopoulos et al (2011) estimated the change of regional climate indices with relevance to agriculture (but not the change of crop yields) In another study (Bank of Greece, 2011), which is so far the only available study of climate change impacts on crops at national level, semi- quantitative (i.e order of magnitude) estimates of future crop yield change are provided These estimates were obtained through the use of crop models for only 3 crops, namely wheat, maize and cotton while for the rest of the crops examined (olive trees, grapevines and vegetables) the estimations were based on findings in international literature published during 1994-2010 and concerning regions other than those of Greece The present study represents a significant addition to the existing knowledge on the impacts of climate change on crops in Greece as it examines a larger range of crops, it provides quantitative estimations of climate change impacts on crop yield change and agricultural income per region and crop Furthermore these estimations are based on models 'tailored' to crop cultivations at different agricultural locations in Greece, capturing in this way the regional/ local dimension of climate change impacts

As indications of the accelerating rate of climate change multiply, including the record breaking mean global temperatures of the last years, and the recent international understandings for mitigation, the built-

in increase will continue and its impacts will be felt at least in the years till 2050 It behooves policy makers to start considering adaptation measures as soon as possible To address this urgent need, in this paper the effects of potential adaptation options on crop yields are also examined, to assess their economic attractiveness and quantify their expected direct economic effects on agricultural income As the crops examined and the conditions expected in Greece are similar to those of neighboring countries such as Italy, Spain, southern France and Cyprus but also some regions of the Balkan peninsular and Turkey, these findings could provide useful insights on applying similar adaptation measures in these regions

The findings may also be of value to enterprises involved in the agricultural sector in both production and in providing focused and innovative financing and insurance In this respect, financial institutions with a considerable fraction of their total exposure being in the agricultural sector, need to ascertain better the corresponding climate change risk To estimate the climate risk of financial institutions, Georgopoulou et al (2014) developed a methodology applicable to many activity sectors The methodology was applied, as a case study, to one of the systemic Greek banks (Piraeus Bank) and the amount at risk was found to be not negligible In view of this finding and considering the fact that the agricultural sector contributes considerably to Greek GDP, the need for adaptation becomes clear, and the choice and effectiveness of measures to be adopted becomes of interest including their cost benefit analysis and fuller coverage of crop varieties and cultivars

The approach for achieving these targets comprises at first the assessment of climate change impacts on important crops under no adaptation by applying an analytical methodology which directly relates climatic parameters with crop yields, and allows the quantitative estimation of impacts in physical terms (% change of crop yield per unit area cultivated) and then in monetary values (change of agricultural income) Next, a number of adaptation options per crop and region are assessed, both in terms of their expected impact on crop yields as well as (to the extent possible) on their private costs and benefits

This section gives an overview of climate change impacts on crops under no adaptation, paying more attention to southern Europe and the period up to 2050 which is the focus of this study

2.1 Cereals

Earlier studies projected a reduction of crop yield for maize in almost all cases, and a small overall increase albeit with considerable regional variation for wheat (e.g Brandao and Pinto, 2002; Trnka et al., 2004; Ministry of the Environment and University of Castilla de la Mancha, 2005; Alexandrov and Eitzinger, 2005; Wiggering et al., 2008) Recent studies confirm the negative effect of climate change on maize (particularly in southern Europe), while for wheat the previously positive impacts are now re-considered (e.g Asseng et al., 2013; Thaler et al., 2012; Kersebaum and Nendel, 2014; Vanuytrecht et al., 2015; Graß et al., 2015; Valverde et al., 2015)

According to a global assessment study (Balkovič et al., 2014) utilizing Representative Concentration Pathways (RCPs) scenarios, wheat yields in 2041-2060 will decrease up to 40% from 2000 in eastern Europe, and change by -8% up to +15% in southern Europe, by -12% up to +5% in northern Europe and

by -10% up to +2% in western Europe (depending on the RCP)

Supit et al (2012) carried out crop simulations covering 35 European countries for SRES scenarios (A2 and B1) Wheat yields in 2030 will increase from 1990-2008 in most countries (Greece: 21-22%, rest southern Europe: 7-13%, other: 0-44%); this trend continues up to 2050 For maize, yields in the Balkans and south-eastern Europe by 2030 decrease or remain stable (Greece: -4%, other: -2% up to -7%), while

by 2050 they will decrease further in southern Europe (Greece: -16%, rest southern countries: -10-16%) Donatelli et al (2012) examined wheat production in the-27 and for the period up to 2030 under the A1B scenario Under the ‘cold’ version of A1B (ECHAM5 data), wheat yields decrease by 5-30% in almost all parts of Spain, Portugal, and Italy, increase by 5-30% in a large part of Greece and Balkans (as well as

2.2 Vegetables, pulses and legumes

Regarding potato, Supit et al (2012) found that in 2030 and under the A2 and B1 scenarios, yields remain stable or increase in almost all countries compared to 1990-2008 levels (Greece: +6-8%, rest southern Europe: +6-17%, other: -4-+17%) In 2050 and under A2, yields remain stable in most countries

or decrease slightly in some (compared to 2030) apart from northern Europe where they increase further Under B1 y,ields slightly increase in the Atlantic coast of western and northern Europe, Italy and Portugal, and remain unchanged or decrease elsewhere (Greece: +6-8%, rest southern Europe: +6-15%, other: -19-+18%) Vanuytrecht et al (2015) found for Belgium an increase by +16-26% in 2031-2050 under the A1B compared to 1981–2010, while an increase by 3-16% was also found for the UK in 2050s depending on water and fertilization rates (Daccache et al., 2011)

On tomato, a recent study covering the Mediterranean region (Saadi et al., 2015) concluded that yields will not change by 2050 under the A1B scenario as tomato is mostly an irrigated crop; however, under mild or severe water stress, relative yield losses by 10-60% were estimated for most of the region For southern Italy, in particular, Ventrella et al (2012b) found that tomato yield will decrease by 10% during 2030-2059

As for other outdoor vegetables and grain legumes, published research for Europe is limited In southern Portugal, under the A2, A1B and B1 emissions scenarios, the yield of grain legumes was found to decrease by 0.6-1.8% in 2011-2040 and by 1.2-3.8% in 2041-2070 compared to 1961-1990 (Valverde et al., 2015)

2.3 Olive trees and grapevines

The link between olive yield, rainfall and CO2 concentration was explored in Viola et al (2014) for Italy; they concluded that (a) under the present CO2 concentration but a lower rainfall the olive yield will decrease, (b) under a stable rainfall but higher CO2 concentration the yield will increase, (c) under the combined effect of an increased CO2 concentration and a reduced rainfall the increase of yield would be much lower than in (b), i.e of the order of 14% Another study on south-eastern Italy concluded that under the A1B scenario the yield of olive trees by 2050 will be by 8-19% lower than the historic (1951-2000) one (Lionello et al., 2014)

In southern Portugal, the yield of rain fed olives under the A2, A1B and B1 scenarios was found to decrease by 4-7.4% in 2011-2040 and by 8-15% in 2041-2070 compared to 1961-1990 (Valverde et al., 2015) Similarly, in Andalucía, Spain, by 2030-2050 a 15-30% rainfall reduction in the fall (combined with a 7%-9% annual reduction) will cause a decrease of yields by 7% and 3.5% by 2030-50 for rain-fed and irrigated olive trees respectively (Ronchail et al., 2014)

Regarding grapevines, modified climatic conditions are expected to have an impact on yields, as well as

on the wine quality by changing the ratio between sugar and acids (Bock et al., 2011; Santos et al., 2011;

Duchêne et al., 2010) As for yields, these were found to decrease by 1.5-2% in 2011-2040 and by 5.4% in 2041-2070 in southern Portugal compared to 1961-1990 under the A2, A1B and B1 emissions scenarios (Valverde et al., 2015) On the contrary, in northern Portugal (Douro Valley), an increase in wine production by about 10% by the end of the 21st century was estimated under the A1B scenario (Santos et al., 2013) For the Apulia region in southern Italy, a decrease of must and wine production by 20-26% in 2021-2050 compared to 1961-1990 was estimated (Lionello et al., 2014), while for the Tuscany region an average decrease of yield by 12% at 0-200m elevations and by 27% at 400-600m elevations by 2100 compared to 1975-2005 was predicted under the A2 and B2 scenarios (Moriondo et al., 2011) One should keep in mind though that grapevine cultivation may start in new areas not cultivated at present because of a thermal deficit

3-3 Materials and methods

3.1 Regional disaggregation and future climate

As climate change impacts on crops may differ significantly between geographical regions, a suitable spatial scale should be chosen for impact assessment In this work, the present division of Greece into administrative regions (Figure 1), with some aggregations performed (i.e ‘Kentriki and Ditiki Makedonia’, ‘Peloponissos and Ditiki Ellada’) was considered suitable as they are broadly representative

of the climatic classification and at the same time correspond to the disaggregation of the national statistics

(Figure 1 is to be inserted here)

The assessment of impacts per crop was performed in regions where the share of regional crop production to the relevant national total exceeds 10% If by this rule the cumulative share to national total was lower than 85%, then more regions were added to the set until the desired percentage was reached In total, 77 cases were modelled (Table 1)

(Table 1 is to be inserted here)

Regarding future climate, this study focuses on short to midterm time horizon, i.e up to 2050 The simulation of the historic (1961-1990) and future (2021-2050) climate in Greece is based on the results

of the regional climate model RACMO2 (developed by the Netherlands Meteorological Service) for the SRES A1B global emissions scenario (Nakicenovic et al 2000) In each region, 1-2 representative (in terms of historical climatic conditions) locations were selected, and for each of them the outputs of the regional climate model were utilized to provide daily values for maximum, mean and minimum temperature, precipitation, relative humidity, wind, and sunshine duration for each year of the climatic periods examined

3.2 Crop modelling for impact assessment

For the assessment of climate change impacts on crops, agronomic simulation models and regression models were utilized Agronomic models simulate in detail all phases of crop growth and are thus more reliable and allow for a quantitative examination of potential adaptation measures The models were adjusted to the regions examined and thus were 'tailored' to the spatial scale selected For crops where

3.2.1 Agronomic simulation models

For the assessment of climate change impacts on cereals, vegetables, legumes, pulses, sunflower, rice and cotton, the Decision Support System for Agrotechnology Transfer (DSSAT) was utilized DSSAT (Hoogenboom et al., 2015; Jones et al., 2003) has been in use for more than 20 years by researchers, policy makers and others in several countries worldwide It includes detailed crop simulation models which cover a large part of main crops cultivated in Greece, and which simulate crop growth, development and yield as a function of the soil-plant-atmosphere dynamics DSSAT also comprises a very rich database of soil types and agronomic experiments for each crop, and allows the introduction of desired crop management schemes (sowing date, irrigation, fertilization, etc.) Thus, this tool allows for a comprehensive simulation and assessment of the impact of climate variability and climate change on crop growth and yield, and for the assessment of potential adaptation options

For grapevines, the VineLOGIC Virtual Vineyard simulation tool developed by the Cooperative Research Centre for Viticulture (CRCV) in Australia was utilized VineLOGIC (Godwin et al., 2002) is a simulation model of grapevine growth and development incorporating a model of the soil water balance and soil salt balance It uses daily meteorological data as inputs and includes simulation drivers such as soil type, row spacing, pruned bud number, variety, and irrigation Thus, VineLOGIC allows for a detailed simulation of the effects on vine growth and yield from water deficits and waterlogging (associated with reduced/ extreme rainfall under future climate) These characteristics make VINELOGIC a particularly useful tool for climate change impact assessments in vineyards and for evaluating appropriate adaptation strategies

Management and cultivation-related input data to the DSSAT include information on planting date, soil characteristics, planting density, row spacing, planting depth, crop variety, irrigation and fertilizer practices, environmental modifications (e.g CO2 atmospheric concentration), organic residue application, chemical application, and harvest management Weather-related input data include latitude

of the weather stations to be used in the simulations, daily values of incoming solar radiation, maximum and minimum air temperature, and precipitation

Regional models were ‘tailored’ to the reality of each geographical region in terms of soil types, management practices, and local climate Regarding soils, three basic categories were considered (loam, clay loam, sandy loam), with different sub-types per region Cultivars were derived from the relevant DSSAT database, with an effort to select those closer to the ones used in Greece Management practices were compiled based on information collected by consultation with agronomists and field visits The ambient CO2 concentration was kept stable at present levels (390 ppm) as its change up to 2030 (the middle of the 2021-2050 period) is not large; in this, the results obtained could be considered somehow 'conservative' as they omit the potential benefits of the CO2 fertilization effect in some cases

high-Historical data on the main climatic parameters (maximum, minimum and average temperature, rainfall, sunshine) from representative meteorological stations in each region were used as input data to the models

The total number of ‘tailored' cases (i.e combinations of soil types, cultivars/varieties, and regions) simulated by agronomic models amounted to 2,042 These were examined under both the historic and the

future climate

As for the models' calibration, to date in Greece a database with agronomic experiments and other relevant information does not exist Thus, to minimize, to the extent possible, the deviation between the models' simulated yields and the historic ones per crop and region, the following two-step 'screening process' was developed:

(1) For each combination of soil type - cultivar per crop and region, the deviation between the simulated crop yield under the historic climate (1961-1990) and the real annual yield (from data published by the National Statistical Service) during 2000-2006 was calculated

(2) For each region and crop, only the combinations with a deviation of 10% of less for at least one year in 2000-2006 were retained for impact assessment, unless all figures were above this limit (and thus all combinations had to be retained)

This approach provided for a rough 'calibration' of agronomic models utilized Table 2 shows the median deviations of the final combinations retained

(Table 2 is to be inserted here)

As seen, the models’ performance can be considered as satisfactory in most of cases, with resulting median deviations being less than ±15% Only six cases have a deviation higher than 30%; however, since it is the difference between the simulated present and future crop yield (and not the absolute figures) that matters for the assessment, this large deviation was not regarded as critical

3.2.2 Regression models

In case of crops for which agronomic simulation models are not available (i.e olive trees, tobacco, orange trees, peach trees, cucumber), annual yields were simulated by linear regression models connecting the crop yield (expressed in tons per ha) with statistically important climatic parameters Μodels were developed on the basis of statistical data on climatic parameters, cultivated areas and production per crop for the time period 1980-2006 Data on climatic parameters for this period derived from the official annual statistical yearbooks of Greece (ESYE 1980-2006(a)), and data on cultivated areas and production per crop derived from the official annual agricultural statistics of Greece (ESYE 1980-2006(b)) As climatic data are available on a monthly basis, climatic parameters in the models are also expressed on the same time basis The models are presented in Table 3

(Table 3 is to be inserted here)

The R and significance parameters (sF) of the models (also shown in Table 3) are a measure of their deviation from real figures on crop yields As seen, in almost all cases the R2-value is equal to or greater than 0.6 while the F-values are small

3.3 Economic evaluation of adaptation measures

The assessment of costs and benefits from the introduction of adaptation measures was done vise-a-vis the 'no adaptation' case where these measures are not implemented Therefore, only the additional cost and benefits from the 'no adaptation' case were considered

The economic evaluation of each adaptation measure was performed separately for each region and crop This was done only where the potential measure was found to reduce yield losses compared to the 'no adaptation' case

The elements included in the evaluation comprised the following:

Cost: purchase and installation of equipment, consulting services for the proper implementation

of the measure, irrigation water supply, fertilizers' supply, etc

Benefits: Decrease of yield losses / increase of yield gains as a result of the measure,

conservation of water for irrigation, etc

The economic evaluation was performed on a unit cultivated area, i.e one Ha

Since the lifetime of each adaptation measure is different, in order to be able to compare the cost and benefits of different measures, capital (investment) costs had to be annualized This was done by applying the equation:

1

, ,

−+

k j k j

r

r r IC

where

i : adaptation measure, j: crop, k: geographical region

ACi,j,k : annualized capital cost of measure i for crop j in region k (€/year)

ICi,j,k : capital cost of measure i for crop j in region k (€)

r: discount rate (%)

T: lifetime of measure (years)

The annual operational and maintenance costs of adaptation measures include the use of any additional irrigation water, the application of additional quantities of chemical N-fertilizers, and rest costs (namely the cost of farmers’ consulting from specialized agronomists on how to properly apply the adaptation measures in field to reduce the adverse effects of climate change)

The equivalent annual cost EACi,j,k were:

k j k

j k j k

j k

j k

j k

CWi,j,k : annual cost of additional irrigation water as a result of measure i for crop j in region k

(€/year)

CFi,j,k : annual cost of additional N-fertilizers as a result of measure i for crop j in region k (€/year)

restOMi,j,k : annual additional rest O&M cost as a result of measure i for crop j in region k (€/year)

The annual benefits Bi,j,k from the implementation of each adaptation measure i for crop j in region k are

given by:

PFN F

F PW W

W P YC

YC

BF BW

BP

B

k j k j NoA k

k j k j NoA j

k j NoA k

j

k j k j k

j i k

j

⋅

−+

⋅

−+

⋅

−

=

=+

+

=

)(

)(

)

, ,

, ,

(3)

where

YC i,j,k : yield of crop j in region k when measure i is implemented (kg/ha)

YC NoA,j,k : yield of crop j in region k under no adaptation (kg/ha)

Pj: producer price of crop j (€/kg)

Wi,j,k : annual consumption of irrigation water for crop j in region k when measure i is implemented

(m3/ha)

WNoA,j,k : annual consumption of irrigation water for crop j in region k under no adaptation (m3/ha)

PWk : price of irrigation water in region k (€/m3)

Fi,j,k : annual consumption of N-fertilizers for crop j in region k when measure i is implemented (kg

less than 1 indicates that measure i is economically attractive for farmers, whereas the opposite (CBR >

1) shows that benefits of the measure are lower than its cost CBR allows comparing adaptation measures which are very different in terms of their lifetime and the magnitude of their costs and benefits

4 Results

4.1 Estimated impacts on crop yields and agricultural income under no adaptation

By applying the models of section 3.3, the percentage estimated change of crop yields between the future (2021-2050) and the historic (1961-1990) climate under no adaptation was calculated (Table 4) The regional figure for each crop simulated by agronomic models corresponds to the median of yield changes estimated for the different combinations of soil types-cultivars retained for this region (see paragraph 3.3.1 above)

(Table 4 is to be inserted here)

Table 4 shows that for some crops a decrease of yield in all regions was estimated (maize, beans, sunflower) Οn the contrary, the future yield of wheat, rice, cotton, orange and peach trees was found to increase In between, one can find:

a) Crops for which the effect of climate change is mostly negative (tomato, pepper, potato, olive trees);

b) Crops which will be mostly benefited from climate change (cabbage, tobacco);

c) Crops with mixed regional effects (barley, grapevine, cucumber)

The yields' changes estimated are in line with the ones estimated by other authors (see Section 2), although some differences are observed in some cases which are probably due to the particular climatic and rest environmental characteristics of the regions examined, as well as to the cultivation practices followed in these regions

By using the cultivated area per crop and region in 2006 (i.e the most recent year for which final data on production and cultivated areas are available from National Statistics), the calculated percentage change

of crop yields in Table 4, and the producer prices of agricultural products in 2012, the expected direct impact of climate change on the annual agricultural income in 2021-2050 was estimated ('Median' case - Table 5) In addition, by utilizing the most adverse future change of yield estimated, the resulting change

of income was also calculated ('Worst' case - also in Table 5)

(Table 5 is to be inserted here)

The addition of economic losses and gains among regions assumes that equal weights are attributed to them regarding their climate vulnerability However, this is not necessarily true as: (a) within a region, one or more crops may be more important in terms of employment or contribution to the regional income; (b) within a crop cultivation, some regions may be more important than others if they have a high contribution to the national production of this crop; and (c) at national level, adverse economic impacts from climate change may be more damaging in some regions than others depending on their adaptive capacity (in terms of infrastructure, specialized personnel, flexibility in substituting cultivations

or changing management practices, etc.) and their dependence on agriculture

Those caveats notwithstanding, the addition of losses and gains provides an indication on the overall vulnerability of agricultural regions and cultivations

By looking at Table 5, the following remarks can be made:

- There are significant differences between crops and regions in terms of economic benefits and losses

- At regional level under the ‘median case’, northern and central Greece and Sterea Ellada & Attiki are climate-winners, while west and southern Greece are climate-losers However, this does not hold in the ‘worst case’, where all regions except Sterea Ellada & Attiki are climate-losers Notably, even in this latter case, regional differences are large

- Cotton (a very water-intensive crop) is the principal reason for climate-winners, and during simulation runs it was assumed that irrigation water supply will continue to be available despite the reduction of precipitation (which will probably have adverse impacts on groundwater replenishment and consequently on the supply of irrigation water) If this assumption does not hold, then from runs performed the results showed that Thessalia and Sterea Ellada join the group of climate-losers, thus leaving only two regions in the north of the country in the group of climate-winners

- At cultivation level, both in the ‘median' and the ‘worst' case, the situation is mixed, with benefits for some cultivations and adverse effects for others

- At national level, the direct losses of the agricultural income in 2021-2050 because of climate change were estimated at about 50-280 million €2012/ year, without considering the potential water stress

effect on water intensive cultivations (i.e cotton and rice) If these cultivations are excluded, then direct losses increase to 160-355 million €2012/ year In relation to the national Gross Domestic Product (GDP), these losses represent 0.08-0.18% of GDP in 2012, a figure which is comparatively low However, regional losses were found to be substantial in some cases, a result obscured in overall national figures of climate change impacts

- Both cases reveal that adaptation is needed at regional level, but regional adaptation efforts are not of equal magnitude and should focus on different cultivations per region

4.2 The effect on crop yields of potential adaptation measures

The following adaptation measures were examined:

terms of adaptation) compared to cultivar breeding and implies a significantly lower cost The magnitude

of benefits on regional crop yields was estimated only for crops simulated by agronomic models taking into account the results for all combinations of soil types-cultivars per region

second by one month later compared to the present planting date The assessment did not aim at determining the optimum planting date and management profile (irrigation, fertilization, etc.) per region and crop

rates was applied (i.e +20% from present levels) as additional amounts of nitrogen will increase water and soil pollution, as well as national N2O emissions limited by the Kyoto Protocol The measure was examined only where simulations under no adaptation showed a nutrients' stress

rates was considered during the periods of water stress, while for rain fed crops (barley, wheat, sunflower) a volume of irrigation water equal to 15% of the monthly future precipitation was assumed to

be available For crops simulated by regression models, only cases where precipitation was found to be statistically significant were examined; in these, during months with a positive correlation of precipitation, irrigation up to 15% of precipitation was assumed to be available (introduced to the model

by means of a 'pseudo'-increase of monthly precipitation)

replacement of existing systems (sprinkler irrigation) with modern ones (micro-irrigation) In contrast with measure M4 where irrigation increased only during periods of water stress under no adaptation, M5 was applied to all irrigation periods

and was examined only for grapevines which are perennial crops (and thus shifting of planting date is not applicable), non-irrigated, and not subject to N-fertilization (as fertilizers may affect the wine’s quality and taste) In this, pruned bud number was reduced, while water supply was left unchanged in order to satisfy the crop’s needs

These six adaptation measures considered represent the main agricultural practices for autonomous adaptation to climate change (Kovats et al, 2014) The advancement or delay of sowing dates can help plants avoid too high or too low temperatures under future climate during critical development stages which can impede growth and thus reduce crop yields in the future The provision of additional quantities

of irrigation water (either through increased irrigation rates or through high-efficiency irrigation which increases the useful amount of water reaching the plants’ roots) represents also a potential adaptation practice, although water availability limitations and sustainability concerns pose a barrier to its wide application Increased N-fertilization may assist plants in filling nutrients’ shortages caused by reduced precipitation under future climate Finally, cultivation and management changes represent the only possible adaptation practice for vineyards when a more radical adaptation strategy (i.e shift of vineyards

to other locations) is not applicable or desired

The impact of adaptation measures on crop yields is presented in Annex A The following remarks apply:

M1: the use of climate-resistant cultivars can have major benefits on crop yields in some cases, limiting significantly yield losses or even increasing yields above present levels

M2: Earlier planting had positive effects on potato, cabbage and (marginally) on sunflower yields, while delayed planting was found to be more promising (although this varies by crop and region) Overall, a shift of planting date seems a promising adaptation measure

M3: In many cases the increase of N-fertilization rates had a minor effect on crop yields For beans (one region) and potato (two regions) there was a small improvement compared to no adaptation Yield losses of pepper and cabbage caused by the lower transportation of nutrients to plants through soil percolation because of reduced precipitation, were counterbalanced by this measure

M4: For maize and beans in almost all regions, and rain fed barley in one region, M4 fully counterbalanced the adverse effects of climate change, while in some other crops/region it resulted at yields even above present levels Yield losses of sunflower were reduced but remained large under M4 For cabbage, higher irrigation had a negative effect as it caused heads of the plant to split and crack Overall, M4 looks an effective adaptation strategy in many cases

M5: This measure was examined only for irrigated crops facing water stress under no adaptation In many cases, M5 was less effective than M4 as it was applied in all irrigation periods (thus, it may provide less water than needed during periods of water stress, and more than necessary in the rest) M5 had a better performance than M4 in cases where the improvement under M4 was marginal

M6: The results confirmed the expected reduction in the crop yield compared to no adaptation; however, the model cannot provide an indication about the effect on the quality of the grapes' wine potential This gap in information does not allow assessing whether lighter pruning will lead to improved wine quality

4.3 Economic attractiveness of potential adaptation measures

Equations (1)-(3) above were applied under a number of assumptions explained below Furthermore, ssensitivity analyses were performed for those cost components which are characterized by a high uncertainty or regional variation These included the unit price of irrigation water, and the producer prices of agricultural products Analyses were not performed for the annual cost of consulting to farmers (which is low and not expected to change significantly during the period considered) As for the capital

cost of high-efficiency irrigation, its components (namely the cost of materials/machinery and the cost of installation) are also not expected to vary significantly in the future

Main assumptions applied were as follows:

Producer prices for crops remain constant at present levels in the base case; however, sensitivity analyses from -30% up to +50% were performed to access the effect of price changes

Prices for irrigation water across Greece vary significantly depending on the geographical region, the origin, quality and quantity of water used, the type of cultivation being irrigated, etc A uniform base value of 0.2 €/m3 was considered (corresponding to the average cost for a farmer when consuming a moderate quality water) Sensitivity analyses up to 4 €/m3 were also performed to exploit the effects of Directive 2000/60/ΕC (‘Water Framework Directive’) which requires that all cost components related to water use must be reflected in water pricing

For grapevines, a cost of variety replacement equal to 10,000 €/ha was considered (lifetime: 20 years) For annual crops, an annual cost of 50 €/ha was assumed for consultancy services in the case

of shift of planting date and cultivar change In addition, a unit cost of 1.3 € per additional kg N in N-fertilizers was assumed The capital cost of new (more efficient) irrigation systems was assumed

to be 6,000 €/ha for grapevines, 3,700 €/ha for vegetables, and 2,400 €/ha for rest crops (including rain fed crops), and for a10-years lifetime

A discount rate of 6% was applied

All utilized cost figures are summarized In Table 6

(Table 6 is to be inserted here)

The calculated Cost-Benefit Ratio (CBR) for each adaptation measure examined and for the base case is displayed in Table 7

(Table 7 is to be inserted here)

The results show that in several cases the change of cultivar and the shift of planting date are the most economically attractive options for cereals under the base value of irrigation water cost (0.2 €/m3) For vegetables, legumes, and pulses, increase of irrigation or N-fertilizers are the measures of first-choice In cases where only one adaptation measure was examined (i.e grapevines and olive trees), its CBR calculated is greater than 1, i.e financial support is needed for farmers to apply this measure For grapevines, the high CBR value is explained by the high replacement cost of existing varieties, and the marginal benefits (in terms of reduced yield losses) considered As for olive trees, the small yield improvement is also the reason behind the high CBR value of increased irrigation

Sensitivity analyses performed on the unit price of irrigation water showed that it affects significantly the economic performance of adaptation measures As expected, at higher prices, high efficiency irrigation improves its CBR and in some cases (maize, beans, tomato, cabbage) it even becomes the first-choice measure (e.g for maize at a price of 0.5-3.5 €/m3 depending on the region) Therefore, the application of the ‘Water Framework Directive’ can have a significant impact on agricultural adaptation On the contrary, increase of irrigation rates is the most attractive option at a price below 0.2 €/m3 in most of cases, while for higher prices its CBR may increase above 1 (i.e costs exceed benefits)

The effect of producer prices on the economic attractiveness of adaptation measures was not found to be particularly significant (at least for the range of price changes examined) in changing the attractiveness order Under prices below the base values considered, the first-choice adaptation option changes only in the case of wheat in one region and of pepper in another

4.4 Direct economic impacts of adaptation at regional and national level

The Net Economic Benefit (NEBi,j,k ) of the adaptation measure i for crop j and region k is the difference

between the annual benefits Bi,j,k and the equivalent annual cost EACi,j,k of equations (2) and (3) above For each crop and region where there was at least one measure with a CBR lower than 1, the NEB of the most economically attractive measure (i.e the one having the lowest CBR) was combined with the relevant cultivated area per crop, and thus the total net economic benefits of adaptation per crop and region were calculated This calculation assumes that all farmers in a region will implement the economically best adaptation measure per crop (i.e a very optimistic approach), supposing that: (a) they have full knowledge of climate change impacts and effects of adaptation, (b) there are no resource-related limitations (e.g restricted availability of irrigation water), (c) there are no other barriers (e.g technical, social, etc.) than cost in applying adaptation

By subtracting the total net economic benefits of adaptation per crop and region from the expected future changes of agricultural income under no adaptation (Table 5 above), the change of annual agricultural income with adaptation was estimated (Table 8) for the ‘median’ and worst ‘cases

(Table 8 is to be inserted here)

At national level, the comparison between Tables 5 and 8 indicates that for the ‘median’ case of climate change impacts, large-scale and cost-efficient adaptation not only outweighs the economic losses to be faced in 2021-2050, but can also bring additional economic benefits by improving yields through low-cost measures, increasing the annual agricultural income by 112 million €/year from present levels, compared to a loss of ca 50 million €/year under no adaptation In the ‘worst’ case, adaptation also reduces significantly the loss of annual agricultural income (from 277 million €/year to 115 million

€/year), but still will not fully mitigate the adverse economic impacts of climate change on crops After

2050 when -based on recent research findings- climate change will be stronger, more ambitious adaptation measures in terms of amplitude and magnitude will be required

One should also not forget that the above ‘optimistic’ conclusions on the net economic gains from adaptation assume also an unlimited availability of irrigation water The penultimate line of Table 8 provides some insight on the importance of this assumption; the annual change of agricultural income remains positive in the ‘median’ case, but is significantly reduced if the economic gain of water-intensive cultivations is not considered

At regional level, in the ‘median’ case under adaptation there are again regional climate-winners and losers (regions in northern-central Greece and southern regions respectively) as is also the case under no adaptation However, if the economic benefits of climate change on cotton are not considered in both cases, then under adaptation, regional winners and losers remain almost the same while under no adaptation all regions become climate-losers In the ‘worst’ case, almost all regions are climate-losers despite large-scale adaptation Pepper and beans are the cultivations (at national level) where adaptation measures would result in reversing the adverse effect of climate change

5 Discussion and conclusions

The analysis carried out has demonstrated that the agricultural sector in Greece will as a whole be affected adversely, albeit with some regional winners This is not surprising in view of the diversity of both the terrain and the diverse climatic conditions found in Greece, which obviously are reflected also in the cultivations and practices found there The economic evaluation of these losses, affecting 18 cultivations covering 60% of the land cultivated (excluding cotton), range from a yearly average of ca

153 million € to a high of 365 million € if the most adverse estimates are taken into account, to be compared with the total output value of the cultivations categories in 2015 of 6,726 million € (EUROSTAT, 2016) Cotton which accounts for 355 million € of value in 2015 (5% of total) is a special case as it seems that the future climatic conditions would lead to an increase of yield resulting in an yearly economic gain of about 87 - 103 Million €, provided that water for irrigation remains available at current levels and prices, something that seems very improbable The same is the case for grapevines and raisin cultivations for which the impact is small (ca 1.4 million € yearly) but with no adaptation measures available that lead to net increases in output value as the ones examined have a high cost

In view of this overall negative effect of climate change in Greece, a number of adaptation measures have been examined, specific to each of the 18 cultivations investigated, and mitigation of damages computed The results show that the use of appropriate adaptation measures may result in a net yearly benefit of about 162 million €

Sensitivity analysis of the adaptation effectiveness as regards the price of water carried out in this work, has shown that an increase of water cost (including all appropriate charges to reflect true value) from a current base value of 0.2 €/m3 has a significant effect on the economic attractiveness of adaptation measures

The results obtained have a number of policy implications At first, the study has shown that there will be regional losers but also winners, with a clear distinction between northern and southern Greece Thus, efforts to address climate change impacts on agriculture need to be tailored to both geography and crop kind even in a comparatively small country such as Greece

In addition, the agriculture of the smaller island regions (Ionian, North and South Aegean) shows a rather small exposure to climate change impacts, mainly because of the small volume of the crops under investigation cultivated there compared to the rest of the regions That is not to say that there is no impact but rather that the economic dimension is small Supporting agriculture in these islands which are main tourist destinations (thus have a large demand of agricultural products especially of the perishable kind such as fruits and vegetables) is important; in the future, this support should be focused on mitigating water stress, which even in today’s conditions is a major disadvantage

Within the context of adaptation, the issue of water stress, especially in the semi-arid Mediterranean area,

is central As shown, the measure of increased irrigation is economically attractive in almost all regions and crops only for low water prices (i.e not exceeding 0.5-1 €/m3), and even then in many cases it is not the first choice of adaptation measures to apply Of more interest is the measure of increased irrigation efficiency, whose attractiveness increases with the water price and in many cases at values over about 1-1.5 €/m3 becomes the most attractive option In view of the fact that the price of water is expected to go

up rather than decrease, the technologies and practices that increase efficiency should be promoted both

in the research and in the development/ demonstration/ dissemination phase